{"id":"W7077883051","doi":"10.48448/yapw-z370","title":"SimSMoE: Toward Efficient Training Mixture of Experts via Solving Representational Collapse","year":2025,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Scalability; Representation (politics); Similarity (geometry); Routing (electronic design automation); Artificial neural network; FLOPS; Constant (computer programming); Training (meteorology)","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002259522,0.001885414,0.001563997,0.0009028169,0.000769039,0.001169257,0.002877419,0.002509082,0.005198982],"category_scores_gemma":[0.0101844,0.001154031,0.001508375,0.0008456877,0.001011051,0.002702054,0.002893457,0.003856039,0.002348504],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001021356,"about_ca_system_score_gemma":0.00224752,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009951676,"about_ca_topic_score_gemma":0.01403994,"domain_scores_codex":[0.9988621,0.000457827,0.00004875489,0.0002725873,0.0001940517,0.0001647176],"domain_scores_gemma":[0.9969113,0.001971097,0.0001653772,0.0002948833,0.0004948783,0.0001624384],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002001861,0.00016026,0.001182215,0.0001234525,0.0001144202,0.0001309621,0.0001452787,0.8102782,0.003146558,0.009286961,0.009834889,0.1653966],"study_design_scores_gemma":[0.000008955036,0.00001372996,0.00002292332,0.000003884265,0.000003974357,0.000008679387,0.000008919334,0.9958178,0.000411795,0.003368979,0.0003270789,0.000003220252],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01545589,0.000271521,0.9792898,0.0003978761,0.00008630892,0.0000826036,0.0001428111,0.002603081,0.001670179],"genre_scores_gemma":[0.3602168,0.0002712986,0.6275252,0.001535808,0.0002181395,0.0004668677,0.001431327,0.0009892707,0.007345333],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.009951676,"threshold_uncertainty_score":0.01978749,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0246899707046746,"score_gpt":0.2786816864818582,"score_spread":0.2539917157771836,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}